面向无参考图像的清晰度评价方法研究

Research of Definition Assessment based on No-reference Digital Image Quality

图像的清晰度是衡量数字图像质量优劣的重要指标。基于面向无参考图像质量评价,重点介绍了目前较为常用的、具有代表性的几种清晰度评价算法,并从算法的单峰性、无偏性、灵敏性,与图像尺度、内容的无关性,以及与主观感知的一致性3个方面的性能进行了对比和分析,以求准确客观地评价各清晰度算法,达到在实际应用中合理、有效地评价图像清晰度的目的。

Image definition is an important indicator of digital image quality.Oriented to the application of No-reference image quality assessment,some representative definition evaluation algorithms are introduced in this paper.In order to accurately and objectively assess each algorithm and to effectively evaluate image definition in the practical application,comparisons and analysis on performance of each algorithm are also done from the following aspects：single peak,no bias,sensitivity;the independence of image size and content;the consistency with subjective perception.Results show the ReBlur and NRSS algorithm have better evaluation performance than other algorithms for the No-reference image quality assessment.